drm/i915: Track has_audio in the pipe config
Including state readout and cross-checking. This allows us to get rid
of crtc->eld_vld on hsw+. It also means that fastboot will be unhappy
if the BIOS hasn't set up the audio routing like we want it too.
Wrt fastboot and external screens I see a few options:
- Don't.
- Try to fix up eld, infoframes and audio settings after the fact. But
that means some pretty extensive reworking of our code which
currently does all this while the pipe/port is still off.
I won't bother with converting SDVO over to this because the audio
support for SDVO is very lacking:
- We don't update the eld.
- We don't update the audio state on the sdvo encoder.
- We don't check whether the platform can even feed audio to the sdvo
encoder.
I've converted hdmi, dp & ddi all in one go since ddi needs both hdmi
and dp converted and so doing it step-by-step would have required a
few intermediate hacks.
Reviewed-by: Naresh Kumar Kachhi <naresh.kumar.kachhi@intel.com>
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
diff --git a/drivers/gpu/drm/i915/intel_drv.h b/drivers/gpu/drm/i915/intel_drv.h
index 1a5ecc9..0ef2777 100644
--- a/drivers/gpu/drm/i915/intel_drv.h
+++ b/drivers/gpu/drm/i915/intel_drv.h
@@ -276,6 +276,10 @@
/* Whether we should send NULL infoframes. Required for audio. */
bool has_hdmi_sink;
+ /* Audio enabled on this pipe. Only valid if either has_hdmi_sink or
+ * has_dp_encoder is set. */
+ bool has_audio;
+
/*
* Enable dithering, used when the selected pipe bpp doesn't match the
* plane bpp.
@@ -366,7 +370,6 @@
*/
bool active;
unsigned long enabled_power_domains;
- bool eld_vld;
bool primary_enabled; /* is the primary plane (partially) visible? */
bool lowfreq_avail;
struct intel_overlay *overlay;